The behavioral variant of Alzheimer's disease (bvAD) is characterized by early and predominant behavioral changes, resembling the clinical profile of the behavioral variant of frontotemporal dementia (bvFTD). Social cognition deficits form hallmark features in bvFTD and altered biometric responses to socioemotional cues have been observed in bvFTD. However, little is known about social cognition and its biometric signature in bvAD. In this explorative study, we investigated all levels of social cognition (i.e., level-1: perception, level-2: interpretation and level-3: reasoning), using the Ekman 60 faces test (level-1), Interpersonal Reactivity Index (IRI) and empathy eliciting videos (level-2), the Social Norms Questionnaire (SNQ) and moral dilemmas (level-3), while measuring eyemovements and galvanic skin response (GSR). We compared 12 patients with bvAD with patients with bvFTD (n=14), typical AD (tAD, n=13) and controls (n=15), using ANCOVAs and post hoc testing, adjusting for age and sex. Regarding perception, bvAD (40.1±8.6) showed lower scores on the Ekman test compared to controls (50.1±4.6, p<0.001), and tAD (46.2±5.3, p=0.05) and higher scores compared to bvFTD (32.4±7.3, p=0.002). Eyetracking during the Ekman test revealed that groups did not differ in dwell time on the eyes (all p>0.05), but bvAD (18.7±9.5%) and bvFTD (19.4±14.3%) spent significantly less dwell time on the mouth when viewing the faces than controls (30.4±10.6%, p<0.05) and tAD (32.7±12.1%, p<0.01). Regarding empathy, bvAD (11.3±4.6) exhibited lower scores on the IRI Perspective Taking subscale compared with controls (15±3.4, p=0.02) and similar scores to bvFTD (8.7±5.6, p=0.19) and tAD (13.0±3.2, p=0.43). The GSR to empathy eliciting videos did not differ between groups (all p>0.05). Regarding knowledge of social norms, bvAD (16.0±1.6) and bvFTD (15.2±2.2) showed lower scores on the SNQ than tAD (17.8±2.1, both p<0.05) and controls (18.1±1.3, both p<0.01). Regarding moral reasoning, no differences among the groups were observed in responses to moral dilemmas (all p>0.05), while only bvFTD (0.9±1.1) showed a lower GSR during the personal condition compared with controls (3.2±3.3 peaks per minute, p=0.02). In conclusion, bvAD showed a similar though milder social cognition profile and a similar eyetracking signature compared with bvFTD and greater social cognition impairments and divergent eyemovement patterns compared with tAD. Our results suggest that bvAD and bvFTD show reduced attention to salient features during facial expression perception, potentially contributing to their emotion recognition deficits. These social cognition and biometric measures provide important insights into the basis of behavioral changes in bvAD, and might be valuable for its clinical diagnosis.